Support read-only duplicates of Mojo shared buffers.

Support read-only duplicates of Mojo shared buffers.

BUG=556587
Committed: https://crrev.com/ee191a43debecee40db66052a2de1621f50a2c49
Cr-Commit-Position: refs/heads/master@{#385634}

[Anand Mistry (off Chromium), 2016-04-05 11:00:10]

amistry@chromium.org changed reviewers:
+ rockot@chromium.org

[Anand Mistry (off Chromium), 2016-04-05 11:00:11]

Take your time. No immediate need for this one. Just fleshing out the remaining
capabilities.

[Ken Rockot(use gerrit already), 2016-04-06 14:37:46]

Cool. Is it hard to support read-only duping in the sync broker here?

https://codereview.chromium.org/1689053003/diff/40001/mojo/edk/system/shared_...
File mojo/edk/system/shared_buffer_unittest.cc (right):

https://codereview.chromium.org/1689053003/diff/40001/mojo/edk/system/shared_...
mojo/edk/system/shared_buffer_unittest.cc:292: // Send a read-only copy of the
buffer to the parent.
nit: "to the child"?

[Anand Mistry (off Chromium), 2016-04-07 00:34:33]

Linux support has a couple of things:
1. Only the process that created the buffer can make a read-only duplicate. I've
updated the API to state this. The other option is to continually pass around
both read-write and read-only fds, which is... messy. Also, it would add
exceptions to the "take shared memory" embedder API.
2. In order to modify the sync broker, I need to modify base::SharedMemory. I
figured doing so should be a separate change.

https://codereview.chromium.org/1689053003/diff/40001/mojo/edk/system/shared_...
File mojo/edk/system/shared_buffer_unittest.cc (right):

https://codereview.chromium.org/1689053003/diff/40001/mojo/edk/system/shared_...
mojo/edk/system/shared_buffer_unittest.cc:292: // Send a read-only copy of the
buffer to the parent.
On 2016/04/06 14:37:46, Ken Rockot wrote:
> nit: "to the child"?

Done.

[Ken Rockot(use gerrit already), 2016-04-07 00:44:00]

lgtm

On 2016/04/07 at 00:34:33, amistry wrote:
> Linux support has a couple of things:
> 1. Only the process that created the buffer can make a read-only duplicate.
I've updated the API to state this. The other option is to continually pass
around both read-write and read-only fds, which is... messy. Also, it would add
exceptions to the "take shared memory" embedder API.

Ah. I don't think this is problematic really. I'd be fine with us making that
constraint explicit, or at least conveying in the public API docs that duping
read-only from a process other than the original creating process is unreliable
and may fail.

> 2. In order to modify the sync broker, I need to modify base::SharedMemory. I
figured doing so should be a separate change.

sgtm

> 
>
https://codereview.chromium.org/1689053003/diff/40001/mojo/edk/system/shared_...
> File mojo/edk/system/shared_buffer_unittest.cc (right):
> 
>
https://codereview.chromium.org/1689053003/diff/40001/mojo/edk/system/shared_...
> mojo/edk/system/shared_buffer_unittest.cc:292: // Send a read-only copy of the
buffer to the parent.
> On 2016/04/06 14:37:46, Ken Rockot wrote:
> > nit: "to the child"?
> 
> Done.

[Anand Mistry (off Chromium), 2016-04-07 01:15:03]

The CQ bit was checked by amistry@chromium.org

[commit-bot: I haz the power, 2016-04-07 01:15:41]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1689053003/60001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1689053003/60001

[commit-bot: I haz the power, 2016-04-07 02:54:19]

Committed patchset #4 (id:60001)

[commit-bot: I haz the power, 2016-04-07 02:56:29]

Description was changed from

==========
Support read-only duplicates of Mojo shared buffers.

BUG=556587
==========

to

==========
Support read-only duplicates of Mojo shared buffers.

BUG=556587

Committed: https://crrev.com/ee191a43debecee40db66052a2de1621f50a2c49
Cr-Commit-Position: refs/heads/master@{#385634}
==========

[commit-bot: I haz the power, 2016-04-07 02:56:30]

Patchset 4 (id:??) landed as
https://crrev.com/ee191a43debecee40db66052a2de1621f50a2c49
Cr-Commit-Position: refs/heads/master@{#385634}